server: health check endpoints are not logged

[enocom]

I configure a server like so:

opts := &server.Options{
    RequestLogger: requestlog.NewNCSALogger(os.Stdout, func(err error) {
        log.Printf("Whoops! Error: %s", err)
    }),
}
s := server.New(opts)
// ... listen and serve, etc.

When I send a request to /, I get a nice log message.

When I send a request to /healthz/liveness or /healthz/readiness, I get no message.

[zombiezen]

This is working as we intended, but perhaps the behavior is surprising. What we're worried about here is log/trace spam from k8s or other orchestrators hitting the health endpoints. They will do this periodically (every few minutes), and this can drown out more legitimate traces and logs, especially for low-traffic servers. Is there something we can do in docs to help surface this?

[enocom]

It might not be the right place, but perhaps in the server.New documentation, a sentence like the following would help:

Note, HealthChecks bypass any configured RequestLogger to prevent flooding the logs. To log a
health check, configure the route as a normal HTTP handler.

[zombiezen]

In reviewing this, let's add a comment to the server docs and we can call this good.

[enocom]

Will do.